IL-35 Suppresses Lipopolysaccharide-Induced Airway Eosinophilia in EBI3-Deficient Mice

Interleukin-35 is a member of the IL-12 family, which is a group of heterodimeric cytokines composed of two of five subunits (p19, p28, p35, p40, and EBI3, IL-12 (p35/p40), IL-23 (p19/p40), IL-27 (p28/EBI3), and IL-35 (p35/EBI3) (1–3). The receptors for the IL-12 family of cytokines are also composed of two of five different subunits: IL-12Rβ1, IL-12Rβ2, IL-23R, WSX-1, and gp130 (1). IL-35R is composed of IL-12Rβ2 and gp130, which are also shared by IL-12R and IL-27R, respectively (1). Although IL-12 and IL-23 are proinflammatory cytokines closely associated with the activation of Th1 and Th17 immune responses, respectively (4–6), IL-27 has IL-12–dependent pro-Th1 and IL-10–dependent immunosuppressive activities (7–9). In contrast, IL-35 is regarded as a purely immunosuppressive cytokine (3, 10, 11).

Although IL-12, IL-23, and IL-27 are mainly produced by APCs, including dendritic cells and macrophages (1, 4–9), IL-35 is primarily produced by IL-35–induced IL-35–producing CD4⁺ regulatory T (iTr35) cells, regulatory B cells, and a subset of plasma cells (3, 10–14). IL-35 is also secreted by a broad spectrum of tissue cells during the course of inflammation, because EBI3 is highly inducible upon inflammatory and other stimuli, whereas p35 is constitutively expressed by various types of cells (15). Upon binding of IL-35 to its receptor, the signals are transduced by the unique STAT1 and STAT4 heterodimer, which induces the expression of target genes, including p35 and EBI3, resulting in further production of IL-35 (16). Thus, IL-35 expands infectious tolerance by regenerating IL-35–producing cells, including iTr35 cells (3, 10, 17). However, the actual role of IL-35 in immunosuppression is still mostly unknown.

Previously, by using an asthma model involving repeated airway exposures to OVA plus LPS and a challenge with OVA, Dokmeci et al. (18) demonstrated that EBI3-deficient mice developed a highly Th2-skewed airway inflammation with increased production of Th2 cytokines (IL-4, IL-5, and IL-13) and high eosinophil numbers compared with wild-type (WT) mice. In the current study, by using a simple LPS-induced acute airway inflammation model, we observed Th2-independent airway eosinophilia in EBI3-deficient mice. Because neither WSX-1–deficient mice nor IL-27p28–deficient mice showed such airway eosinophilia, the lack of IL-35 was likely to be responsible for the enhanced eosinophil influx. In vitro, rIL-35 suppressed production of the eosinophil-attracting chemokines eotaxin-1/CCL11 and eotaxin-2/CCL24 by human lung epithelial cell lines stimulated with TNF-α and IL-1β. In vivo, rIL-35 dramatically reduced LPS-induced airway eosinophil influx with concomitant reduction of CCL11 and CCL24 production in EBI3-deficient mice, whereas neutralization of IL-35 significantly increased airway eosinophils in LPS-treated WT mice. Thus, IL-35 functions as a potent inhibitor of LPS-induced airway eosinophilia, at least in part by reducing CCL11 and CCL24 production.

LPS from Escherichia coli O111:B4 was purchased from Sigma-Aldrich (St Louis, MO). FITC-labeled anti-mouse Ly-6G/Ly-6C (Gr-1; RB6-8C5) and PE/Cy5-labeled anti-mouse CD11c were purchased from BioLegend (San Diego, CA). PE-labeled rat anti-mouse Siglec-F (E50-2440) was purchased from BD Pharmingen (Franklin Lakes, NJ). Rabbit polyclonal anti-mouse EBI3 (bs-8606R), anti-mouse eosinophil peroxidase (EPO) (bs-2343R), anti-mouse CCL11 (bs-1601R), and anti-mouse CCL24 (bs-2483R) were purchased from Bioss (Woburn, MA). A mouse monoclonal anti-mouse IL-35 was purchased from Rockland Immunochemicals (Limerick, PA) and dialyzed against PBS to remove sodium azide before use. A mouse monoclonal IgG1 was purchased from eBioscience (San Diego, CA). The following rabbit polyclonal and mAbs were all purchased from Cell Signaling Technology (Danvers, MA): anti-STAT1, anti-STAT3 (mAb, 79D7), anti-pSTAT1 (Y701) (mAb, D4A7), anti-pSTAT2 (Y690), anti-pSTAT3 (Y705) (mAb, D3A7), anti-pSTAT4 (Y693) (mAb, D2E4), anti-pSTAT5 (Y694) (mAb, D47E7), anti-pSTAT6 (Y641), anti–suppressor of cytokine signaling (SOCS)1 (A156), anti-SOCS2, anti-SOCS3 (L210), and HRP-linked anti-rabbit IgG. Recombinant human TNF-α was purchased from Wako (Osaka, Japan). Recombinant human IL-1β was purchased from R&D Systems (Minneapolis, MN). Recombinant mouse IL-35–human Fc chimeric protein was purchased from Chimerigen Laboratories (San Diego, CA).

C57BL/6 mice were purchased from Japan SLC (Hamamatsu, Japan). The generation of mice deficient for genes encoding EBI3, p28, and WSX-1 was described previously (8, 19, 20). The mice were backcrossed with C57BL/6 mice for at least nine generations to homogenize the genetic background. Mice were housed in a specific pathogen–free facility at the Kindai University Faculty of Medicine. Female mice (8–12 wk old) were used in the experiments. Mice were sensitized and challenged with OVA as follows. Mice were inoculated i.p. with 200 μg OVA in 500 μl PBS mixed with an equal volume of alum adjuvant (Wako) on days 0 and 7. From day 14, mice were briefly anesthetized with Somnopentyl (Kyoritsu Seiyaku, Tokyo, Japan) and inoculated intratracheally (i.t.) with 100 μg of OVA in 50 μl of PBS every day for 4 d. On day 18, bronchoalveolar lavage fluid (BALF) and lungs were obtained. For LPS stimulation, mice were anesthetized briefly with Somnopentyl and inoculated i.t. with 50 μl of PBS alone (mock treated) or PBS containing 200 ng of LPS. In some experiments, we added 2 μg of recombinant mouse IL-35–human Fc chimeric protein, 5 μg of monoclonal anti-mouse IL-35, or 5 μg of control mouse IgG1 to the inoculation, as indicated. After 1 d, BALF and lungs were obtained. To obtain BALF, mice were anesthetized and incised along the cervical region to cannulate the trachea. BALF was collected by washing with 1 ml of 0.1% BSA in PBS five times. Subsequently, the right lungs were removed, fixed in 4% paraformaldehyde, and embedded in paraffin for histology. The left lungs were frozen in liquid nitrogen for RNA extraction. All samples were stored at −80°C until use. Unless otherwise stated, four to six mice were used for each group in individual experiments. This study was approved by the Experimental Animal Care Committee of Kindai University, and all animal experiments were performed in accordance with the institutional guidelines.

Cells in BALF were pelleted and washed in RPMI 1640 containing 2% FBS. Bone marrow cells were removed from femurs by flushing with PBS using a 27-gauge needle and washed with RPMI 1640 containing 2% FBS. WBCs were prepared from peripheral blood by lysing RBCs and were washed with RPMI 1640 containing 2% FBS. To prepare single cells from whole lungs, lung tissues were minced in RPMI 1640 containing 4 mg/ml collagenase D, 10 U/ml DNase I, and 1 U/ml heparin and incubated at 37°C for 40 min with constant shaking. Freed cells were washed with RPMI 1640 containing 2% FBS and filtered using 40-μm nylon meshes (BD Biosciences, San Jose, CA). For flow cytometry, cells were suspended in RPMI 1640 containing 2% FBS and counted on a Muse cell analyzer (EMD Millipore, Billerica, MA). Cells were suspended at 1–5 × 10⁶ cells per milliliter, stained with appropriate fluorochrome-labeled mAb (all from BioLegend), and analyzed by flow cytometry using an LSR Fortessa X-20 cell analyzer (BD) and FlowJo software (TreeStar, Ashland, OR). Neutrophils were defined as Gr-1⁺CD11c⁻, and eosinophils were defined as Siglec-F⁺CD11c⁻.

The following ELISA kits were used: Quantikine ELISA Kits for mouse eotaxin/CCL11 and eotaxin-2/CCL24 (R&D Systems), Mini ELISA Development Kits for mouse CCL5/RANTES and CCL12/MCP-5 (PeproTech, Rocky Hill, NJ), and Mouse IL-35 Heterodimer ELISA Kit (BioLegend).

Lung tissues were fixed with paraformaldehyde and embedded in paraffin. Tissue sections were deparaffinized, treated with citrate buffer (pH 6.0), and incubated with anti-mouse eotaxin-1/CCL11, anti-mouse eotaxin-2/CCL24, anti-mouse EPO, or anti-mouse EBI3 and with a Histofine SAB-PO kit (Nichirei Biosciences; Tokyo, Japan), following the manufacturer’s instructions. Images were recorded at ×200 magnification on a BZ-8000 (Keyence, Osaka, Japan).

Total RNAs were isolated from cell lines and tissues using an RNeasy Isolation Kit (QIAGEN, Tokyo, Japan) and on-column DNA digestion by RNase-free DNase, according to the manufacturer’s protocols. cDNAs were synthesized using a Superscript First Synthesis System for RT-PCR (Life Technologies, Carlsbad, CA) using oligo(dT) primer. Conventional RT-PCR was performed using KOD FX (Toyobo, Osaka, Japan) and the following primer pairs: 5′-CCTGCACCACCAACTGCTTAG-3′ and 5′-GTGGATGCAGGGATGATGTTC-3′ for mouse GAPDH; 5′-CTGAAACAGCTCTCGTGGCTCTA-3′ and 5′-GAGGGTCCGGCTTGATGATT-3′ for mouse EBI3; 5′-ATGTCCACAGCTTTGCTGAATCT-3′ and 5′-CTGCAGCCAGCACCTGAAAG-3′ for mouse p28; 5′-CCGGTCCAGCATGTGTCAA-3′ and 5′-CAGGTTTCGGGACTGGCTAAGA-3′ for mouse p35; 5′-ACTCACATCTGCTGCTCCACAAG-3′ and 5′-CACGTGAACCGTCCGGAGTA-3′ for mouse IL-12p40; 5′-GCCAAGGTCATCCATGACAACTTTGG-3′ and 5′-GCCTGCTTCACCACCTTCTTGATGTC-3′ for human GAPDH; 5′-CCTCTCACGCCAAAGCTCACACCTTC-3′ and 5′-CGGCACAGATATCCTTGGCCAGTTTG-3′ for human CCL11; 5′-CCTTCTGTTCCTTGGTGTCTGTG-3′ and 5′-TTCATGTACCTCTGGACCCACTC-3′ for human CCL24; 5′-TCTGGGAGTGCTGTTCTGCTT-3′ and 5′-TGTGCCTTGGAGGAGTGTGA-3′ for human gp130; and 5′-GAGACTCGACAGCACAACCTGA-3′ and 5′-CTGTAGGCTGCTTATTGGATGTGA-3′ for human IL-12Rβ2.

For relative quantitation of gene expression, we used the 2^−ΔΔC_T method, with GAPDH as the internal control for normalization (21). Real-time PCR was performed on a StepOnePlus (Life Technologies). We used Thunderbird Probe qPCR Mix (Toyobo) and the following TaqMan assay reagents from Applied Biosystems (Foster City, CA): Mm99999915_g1 for mouse and human GAPDH, Mm00445259_m1 for mouse IL-4, Mm00461162_m1 for mouse IL-27p28, Mm00469294_m1 for mouse EBI3, Mm01302427_m1 for mouse CCL5, and Mm01617100_m1 for mouse CCL12. We also used Thunderbird SYBR qPCR Mix (Toyobo) and the following homemade primer pairs: 5′-TCCCATGAGCACAGTGGTGAAAG-3′ and 5′-CACAGTACCCCCACGGACAGTTT-3′ for mouse IL-5; 5′-CATGGCGCTCTGGGTGACTG-3′ and 5′-CGGCCAGGTCCACACTCCATA-3′ for mouse IL-13; 5′-GAGGATCTCTGCCACGCTTC-3′ and 5′-CCTCGACCCACTTCTGATGG-3′ for mouse CCL7/MCP-3; 5′-GCAGTCTGAAGGCACAGCAA-3′ and 5′-GGTTGGCACAGACCTGGAAC-3′ for mouse CCL9/MIP-1γ; 5′-ATCTGTCTCCCTCCACCATG-3′ and 5′-CCCTCAGAGCACGTCTTAGG-3′ for mouse CCL11; and 5′-GCCTCCTTCTCCTGGTAGCC-3′ and 5′-ATGGCCCTTCTTGGTGATGA-3′ for mouse CCL24 and above mentioned human CCL11 and CCL24 primer pairs.

Cells were washed with PBS and lysed in CelLytic M (Sigma-Aldrich) containing Protease Inhibitor Cocktail Complete (Roche Diagnostics, Mannheim, Germany) and Phosphatase Inhibitor Cocktail (Toyobo). After incubation at room temperature for 15 min, cell debris was removed by centrifugation. Cell lysates were electrophoresed on an SDS polyacrylamide gel in reducing conditions and electrophoretically transferred to a polyvinylidene difluoride membrane. Membranes were blocked in 5% skim milk or 5% BSA and probed with primary Abs to STAT1, STAT3, p-STAT1, p-STAT2, p-STAT3, p-STAT4, p-STAT5, p-STAT6, and SOCS3, following the manufacturer’s instructions. After washing, membranes were incubated with HRP-conjugated secondary Abs and developed using an ECL Prime System (GE Healthcare, Chicago, IL). Chemical luminescence was observed on an ImageQuant LAS 4010 (GE Healthcare). Signal intensities were quantified using ImageJ software (National Institutes of Health, Bethesda, MD).

BEAS-2B is a human bronchial epithelial cell line transformed by hybrid adenovirus SV-40 (22), whereas A549 is a human lung adenocarcinoma cell line with the alveolar type II cell phenotype (23). Both cell lines were obtained from American Type Culture Collection. BEAS-2B was maintained in DMEM-F12 (1:1) supplemented with 10% FBS, penicillin (100 U/ml), and streptomycin (100 μg/ml). A549 was maintained in DMEM supplemented with 10% FBS, penicillin (100 U/ml), streptomycin (100 μg/ml), and 2 mM l-glutamine. Cells were seeded at 1 × 10⁵ cells in 0.5 ml per well in 12-well plates, grown to semiconfluence, washed with PBS, kept in serum-free medium for 1 d, and treated with 10 ng/ml of TNF-α and 0.1 ng/ml of IL-1β. In some experiments, recombinant mouse IL-35–human Fc chimeric protein was added at 0.1 ng/ml. After treatment, cells and supernatants were used for RT-PCR, ELISA, and immunoblot.

Statistical analyses were performed using a two-tailed Student t test for two groups and one-way ANOVA with Fisher’s protected least significant difference test for comparison among three or more groups. The p values < 0.05 were considered statistically significant.

Initially, we used two experimental models to examine the role of EBI3 in airway inflammation (Fig. 1A). In the first model, we treated WT and EBI3-deficient mice with i.p. injection of OVA in alum adjuvant twice and then with i.t. inoculation of OVA in PBS four times. BALF was collected 1 d after the last OVA inoculation. As shown in Fig. 1B, total cells and eosinophils were highly increased in BALF of EBI3-deficient mice compared with WT mice. The results were consistent with the highly Th2-skewed airway response of EBI3-deficient mice described by Dokmeci et al. (18). In the second model, we treated WT and EBI3-deficient mice with a single i.t. inoculation of LPS and obtained BALF after 24 h. Although the numbers of total cells and neutrophils in BALF were similar between WT and EBI3-deficient mice, eosinophils were highly enriched in BALF of EBI3-deficient mice (Fig. 1B). Immunohistochemical staining of EPO also confirmed enhanced infiltration of eosinophils in lung tissues and heavy deposits of EPO on the bronchial epithelial surface in LPS-treated EBI3-deficient mice (Fig. 1C).

Because the LPS-induced airway inflammation was highly acute, the classical Th2 responses were unlikely to be responsible for the enhanced eosinophil recruitment seen in EBI3-deficient mice. As shown in Fig. 2, although we confirmed a highly elevated expression of Th2 cytokines in EBI3-deficient mice compared with WT mice by the OVA model, we did not observe any significant expression of Th2 cytokines in lung tissues of WT or EBI3-deficient mice by the LPS acute inflammation model. There also was no difference with regard to eosinophil numbers in the bone marrow, peripheral blood, or whole lung (data not shown) between WT mice and EBI3-deficient mice. Thus, the increased airway eosinophilia in LPS-treated EBI3-deficient mice was likely caused by enhanced local recruitment of eosinophils.

Eosinophils express CCR3 and migrate to the chemokines signaling via CCR3, such as CCL5/RANTES, CCL7/MCP-3, CCL9/MIP-1γ, CCL11/eotaxin-1, CCL12/MCP-5, and CCL24/eotaxin-2 (24). The above results suggested an increased production of some of these chemokines in the airway of LPS-stimulated EBI3-deficient mice. Thus, we examined mRNA expression of these chemokines in lung tissues of WT and EBI3-deficient mice. As shown in Fig. 3A, EBI3-deficient mice upregulated CCL5, CCL11, CCL12, and CCL24 transcripts at much higher levels than did WT mice after LPS stimulation. BALF of EBI3-deficient mice also contained CCL11 and CCL24 at much higher levels than did that of WT mice (Fig. 3B). Rather unexpectedly, however, we did not observe significant increases in CCL5 and CCL12 in BALF of EBI3-deficient mice; these chemokines are not CCR3 specific (24), so we focused on CCL11 and CCL24.

By immunohistochemistry, bronchial epithelial cells were mainly positive for CCL11, and its intensity and airway surface deposition were highly increased in EBI3-deficient mice upon LPS stimulation (Fig. 3C). Furthermore, cells positive for CCL11 were increased in the submucosa of LPS-treated EBI3-deficient mice. As for CCL24, we found that a fraction of alveolar epithelial cells and alveolar macrophages were positive for CCL24, and its signals were highly increased upon LPS stimulation, especially in EBI3-deficient mice (Fig. 3C). These results suggested that the elevated production of CCL11 and CCL24 by airway epithelial cells and alveolar macrophages caused enhanced influx of eosinophils in LPS-stimulated airway of EBI3-deficient mice.

EBI3 is known to pair with p28 and p35 to function as IL-27 and IL-35, respectively (1). Although p28 is specific for IL-27, p35 is also a component of IL-12 (through pairing with p40) (1). We examined mRNA expression of these IL-12 family subunits in the lung tissues of WT and EBI3-deficient mice by semiquantitative RT-PCR (Fig. 4A). By LPS stimulation, EBI3 expression was strongly upregulated in WT mice and p28 expression was upregulated in both WT and EBI3-deficient mice. p35 expression was constitutive in WT and EBI3-deficient mice, and p40 expression was not detected.

Thus, the enhanced airway influx of eosinophils in EBI3-deficient mice could be due to the lack of IL-27 (p28/EBI3), IL-35 (p35/EBI3), or both. To test these possibilities, we used p28-deficient mice and WSX-1/IL-27R α-subunit–deficient mice. As shown in Fig. 4B, BALF of all groups of mice showed similar total cell numbers and neutrophil numbers, but only BALF of EBI3-deficient mice showed high eosinophil numbers. Thus, the lack of IL-35, but not IL-27, was likely responsible for the enhanced airway influx of eosinophils. By immunohistochemistry, EBI3 was detected mainly in alveolar and interstitial macrophages of LPS-stimulated WT mice (Fig. 4C). We also confirmed that IL-35 was highly increased in BALF from LPS-stimulated WT mice (Fig. 4D).

We next examined the effect of rIL-35 on CCL11 and CCL24 production by human lung epithelial cell lines BEAS-2B and A549. Consistent with previous reports (25, 26), treatment with TNF-α and IL-1β induced CCL11 expression in BEAS-2B cells, as well as CCL11 and CCL24 expression in A549 cells (Fig. 5A). We also found that both cell lines expressed IL-12Rβ2 and gp130, the heterodimeric subunits of IL-35R (Fig. 5A). As shown in Fig. 5B, rIL-35 potently suppressed the upregulation of CCL11 and CCL24 induced by TNF-α and IL-1β at the mRNA (left panels) and protein (right panels) levels in A549 cells. Similarly, IL-35 potently suppressed the upregulation of CCL11 in BEAS-2B cells treated with TNF-α and IL-1β at the mRNA (left panel) and protein (right panel) levels (Fig. 5C). These results demonstrate that IL-35 functions as a potent inhibitor of CCL11 and CCL24 expression in human lung epithelial cell lines stimulated with proinflammatory cytokines.

We further examined the effect of IL-35 on the phosphorylation of STAT proteins and expression of SOCS proteins in BEAS-2B cells. As shown in Fig. 5D, IL-35 inhibited the phosphorylation of STAT1 and STAT3 in BEAS-2B cells stimulated with TNF-α and IL-1β. Furthermore, IL-35 increased SOCS3 expression, irrespective of the stimulation with TNF-α and IL-1β. These results support that IL-35 exerts a potent regulatory effect on the signaling pathways of lung epithelial cells.

To determine the in vivo role of IL-35, we examined the effect of i.t. inoculation of rIL-35 or anti–IL-35 on LPS-induced airway eosinophilia. As shown in Fig. 6A, IL-35 did not affect the number of total cells or neutrophils (left panel) but dramatically reduced eosinophil numbers (right panel) in BALF of LPS-stimulated EBI3-deficient mice. Furthermore, the airway delivery of IL-35 significantly reduced the production of CCL11 and CCL24 in BALF of LPS-stimulated EBI3-deficient mice (Fig. 6B). In contrast, the airway delivery of anti–IL-35 did not affect the numbers of total cells or neutrophils (left panel) but significantly increased eosinophil numbers (right panel) in BALF of LPS-stimulated WT mice (Fig. 6C). Collectively, these results support that IL-35 suppresses LPS-induced airway eosinophilia, at least in part by reducing the production of CCL11 and CCL24.

In this study, we demonstrated a highly exaggerated airway eosinophilia in EBI3-deficient mice not only by the standard OVA asthma model but also by a much simpler LPS-induced acute airway inflammation model (Fig. 1). Because the LPS-induced airway inflammation peaked by 1 d, Th2 immune responses could not be involved in the LPS-induced airway inflammation. Indeed, although a highly elevated expression of Th2 cytokines was observed in the lung of OVA-treated EBI3-deficient mice, Th2 cytokines were barely detected in the lung of LPS-treated EBI3-deficient mice (Fig. 2). EBI3-deficient mice also do not have systemic increases in eosinophils in the bone marrow, blood, or lung compared with WT mice (data not shown). Thus, it was considered that the local recruitment of eosinophils was greatly enhanced in the LPS-stimulated airway of EBI3-deficient mice. Indeed, we found that EBI3-deficient mice had a large increase in the production of eosinophil-attracting chemokines CCL11 and CCL24 in BALF upon LPS stimulation (Fig. 3). Because IL-27p28–deficient mice and WSX-1/IL-27Rα–deficient mice did not have such increases in airway eosinophils upon LPS stimulation (Fig. 4), we considered that the enhanced eosinophil influx and the elevated production of CCL11 and CCL24 in the LPS-stimulated airway of EBI3-deficient mice were primarily due to the lack of IL-35.

Indeed, we demonstrated that rIL-35 potently suppressed production of CCL11 and CCL24 by human lung epithelial cell lines treated with TNF-α and IL-1β (Fig. 5). We also demonstrated that airway delivery of rIL-35 dramatically reduced eosinophil numbers and production of CCL11 and CCL24 in BALF of LPS-stimulated EBI3-deficient mice, whereas that of anti–IL-35 significantly increased eosinophil numbers in BALF of LPS-stimulated WT mice (Fig. 6). Thus, our findings demonstrate that IL-35 suppresses LPS-induced airway eosinophilia, at least in part by reducing local production of CCL11 and CCL24. However, to exclude alternative mechanisms, future experiments, such as those using EBI3-deficient and CCL11/CCL24 double-deficient mice and/or CCL11/CCL24-transgenic mice, will be required.

By using BEAS-2B cells, we also demonstrated that IL-35 suppressed phosphorylation of STAT1 and STAT3 induced by TNF-α and IL-1β (Fig. 5). Of note, STAT3 was shown to be involved in CCL11 expression (27, 28). We also found that IL-35 increased the expression of SOCS3 in BEAS-2B cells, irrespective of stimulation with TNF-α and IL-1β (Fig. 5). SOCS3 is known to inhibit IL-6 signaling through interaction with gp130 (29). The results support that IL-35 has a potent regulatory effect on the signaling pathways of airway epithelial cells.

A number of studies demonstrated enhanced immune responses in EBI3-deficient mice by using various disease models. A possible caveat of such studies is that EBI3 functions as a subunit of two immunomodulatory cytokines, IL-27 and IL-35 (1); thus, the observed changes might be due to the lack of either or both of them. For example, Tong et al. (30) reported that EBI3-deficient mice had aggravated delayed-type hypersensitivity reactions, with increased production of IL-17 and reduced production of IL-10. IL-27 and IL-35 are known to suppress the development of Th17 and production of IL-17, whereas IL-27 also induces IL-10 production in effector T cells (8, 31). Thus, the results could be mainly explained by the lack of IL-27, although the lack of IL-35 may also be partly responsible. Similarly, Tirotta et al. (32) demonstrated that infection of EBI3-deficient mice with a neuro-adapted mouse hepatitis virus resulted in increased mortality, with enhanced infiltration of T cells and macrophages into the CNS. Increased secretion of IFN-γ and decreased secretion of IL-10 by virus-specific CD4⁺ and CD8⁺ T cells were observed in EBI3-deficient mice compared with WT mice (32). Again, the results could be mainly explained by the lack of IL-27 (8). In contrast, Wirtz et al. (33) compared EBI3-deficient and IL-27p28–deficient mice in various colitis models. Although p28-deficient mice were phenotypically similar to WT mice, EBI3-deficient mice developed aggravated intestinal diseases. Furthermore, administration of rIL-35 effectively suppressed colitis induced by dextran sulfate sodium or trinitrobenzene sulfonic acid in EBI3-deficient mice (33). Thus, the lack of IL-35, rather than that of IL-27, may be responsible for the aggravated intestinal inflammation in EBI3-deficient mice.

Using a protocol of airway sensitization with OVA and LPS and challenge with OVA, Dokmeci et al. (18) also demonstrated that EBI3-deficient mice developed a highly aggravated Th2-type airway inflammation, with highly elevated Th2 cytokines and enhanced airway eosinophilia compared with WT mice. The investigators speculated that EBI3 deficiency primarily led to a decrease in Th1 differentiation, which, in turn, led to enhanced Th2 responses in EBI3-deficient mice (18). They did not examine the role of IL-27 or IL-35 in their asthma model. Furthermore, because they did not find enhanced Th2 responses by the epicutaneous route of OVA exposure in EBI3-deficient mice (18), a highly Th2-skewed immune response in EBI3-deficient mice may be organ specific. Based on our findings, we speculate that an enhanced production of eosinophil-attracting chemokines (CCL11 and CCL24) in the absence of IL-35 may be partly responsible for the highly enhanced airway eosinophilia in EBI3-deficient mice. Consequently, it would be of great interest to test whether airway delivery of IL-35 also significantly reduces airway eosinophilia in the OVA asthma model.

Recently, also by using the OVA asthma model, Dong et al. (34) reported that i.p. injection of IL-35 during the allergen-sensitization stage efficiently ameliorated airway hyperresponsiveness, with significant reductions in Th2 cytokines and eosinophil numbers in BALF. They demonstrated that the accumulation of dendritic cells in the mediastinal lymph nodes and lung was significantly reduced by IL-35 treatment. However, they found no direct effect of IL-35 on the surface phenotypes or chemokine production of dendritic cells (34). Thus, the effect of IL-35 appeared to be on tissue cells. Based on our findings, we speculate that a reduced production of some chemokines attracting dendritic cells by IL-35 may be partly responsible for the observed reduction in the recruitment of dendritic cells in the mediastinal lymph nodes and lung.

In conclusion, our findings suggest that IL-35 negatively regulates airway eosinophilia, at least in part by reducing the production of eosinophil-attracting chemokines CCL11 and CCL24. Given that increased accumulation of eosinophils is known to be pathogenic in various diseases, such as allergic asthma, eosinophilic pneumonias, and eosinophil-associated gastrointestinal disorders (35–39), IL-35 may provide a new therapeutic tool to reduce tissue recruitment of eosinophils in such diseases. Furthermore, the suppressive effect of IL-35 on stimulus-induced production of chemokines may not be limited to CCL11 and CCL24; it may well extend to other chemokines targeting different types of leukocyte. This possibility remains to be explored.

The authors have no financial conflicts of interest.